DeutschInternal network
Your own CA, push and outside calendars off: only email goes outside.
Public sector
Digital sovereignty you can operate yourself
Public bodies need control over data, software and AI. Xenaxio runs entirely on your servers – with your own AI model, open standards and an open-source web client.
What a day looks like
AI on your hardware
An open model with Ollama or vLLM in your data centre.
Citizens book appointments
Online booking on your domain, without an account.
What matters to you
- Open standardsIMAP, SMTP, CalDAV, CardDAV, Matrix.
- Open source clientWeb client under AGPL-3.0.
- Audit logEvery AI action traceable.
- Chat with other bodiesOver Matrix, only with servers your admin allows.
In detail
Running Xenaxio in your own data centre
On your own servers, Xenaxio is the same software as the cloud, with every feature. As a guide: up to 100 people need 8 vCPU, 16 GB RAM and 160 GB SSD plus mailboxes; for the AI, a model of the 24-billion class (for example Mistral Small) on a GPU with 24 GB, run with Ollama or vLLM. Updates come when you choose; nothing updates itself.
Nothing goes outside except email
Everything that would leave the network can be switched off: the AI runs on your model instead of Mistral, certificates come from your own CA instead of Let’s Encrypt, push notifications and outside calendars are off, and Rooms talks only to the servers you allow. Sending and receiving email naturally needs other mail servers.
Open standards, citizens and other bodies
Data stays in open formats – IMAP, SMTP, CalDAV, CardDAV, Matrix – and the web client is open source under AGPL-3.0, so you are not locked in. Citizens book appointments on your domain without an account and get a reminder the day before. Video meetings run on your own media server; webinars for information events with up to 300 viewers are available on request.
How to start
Try the live demo without signing up: for an hour you are the administrator of a complete sample company. For our first customers we run a pilot programme – we set up Xenaxio personally and move your mailboxes – and you get a quote, usually within one business day.
More industries: Law firms · Tax advisers · Consultancies · Agencies · Ship management · Mid-sized companies
Questions
Can we run Xenaxio on our own servers?
Yes, Xenaxio is built for that from the start: your data centre runs the same software as our cloud, with every feature. With your own AI model, for example through Ollama or vLLM, AI requests stay in your network too. We help with installation and updates – just write to us.
Which AI does Xenaxio use?
By default, language models by Mistral AI through its EU endpoint. Your administrator can connect your own model instead – for example Ollama or vLLM on your own hardware, wherever you run it. Content then goes only there. Your administrator decides which model your team uses and who may use the AI.
Is Xenaxio open source?
The web client is: it is based on Bulwark Mail and published under AGPL-3.0 on GitHub. We run the platform behind it – administration, AI and the connection to AI chats – for you.
Where is our data stored?
Mailboxes, files, calendars, contacts and the database live on servers in Germany. AI processing runs through the EU endpoint of Mistral AI, a French provider – or, if your company connects its own model, only on your own server.
Ready for a sovereign workspace?
In our cloud or on your servers – we set up Xenaxio for your team.